[Openchrome-users] vt1625 and PAL modes?

[Ken Huisman]

On Fri, 2007-02-02 at 10:25 +0100, Xavier Bachelot wrote:
> Ken Huisman wrote:
> 
> > For what it is worth, I am no longer making any changes for VT1625 - the
> > patch available is probably the last one you will see from me - a couple
> > months ago I did something to hose my commell board so I don't have a
> > test platform - it would be nice to submit the patch I have into the
> > source stream, it is better than what was there before, although not
> > perfect.
> >
> OK, I also think it's time to commit this, so let's summarize.
> 
> Your patch adds :
> - 720x480P : this is progressive ? What is this used for ?

this is what was 720x480Under before my changes, which apparently worked
as a progressive mode resolution, so I renamed it.

> - 720x576Over PAL
> - 640x480 PAL
> 
> It also modifies :
> - 720x480Under NTSC
> - 720x480Fit NTSC
> - 720x480Over NTSC
> 
> > Eventually I may get back to trying to resurrect my commell board (I'm
> > hoping I just need an OS reinstall), but its use in my home has been
> > replaced, so the board may just get sold.  Its too bad, I was really
> > close to getting NTSC 720x480Over working.
> > 
> Do you think the improvements to this mode worth being committed or not ?

Brace yourself, here is a long tale!
its a lot closer to working than it was, but it is not usable yet.  I
think all the VT1625 register settings are correct, but that the
modeline info in the driver needs to be changed.
To this email, I'm attaching a newer diff than the one on my website.  I
don't know the status of this diff (since I don't have any hardware to
test with), the PAL modes should be exactly the same as before, this
patch contains some mucking around with the modelines for 720x480Under
and 720x480Over (plus some more register changes for the latter).  I
recall getting fairly close to having 720x480Over working (even at 24
bit color, which was my focus) - If I recall correctly, I got it close
to looking like what a real overscan mode was supposed to look like, but
it was a bit more flickery than it should have been, and I hadn't been
able to figure that out by the time that i busted my commell board.  So
anyways, what is in the attached diff was a work in progress, but could
be tested by someone - it might be better than what was in the diff on
my webpage.  It would be worth trying to fire up this patch in 24 bit
color on the 480Under and 480Over resolutions and see if they work at
all, I can't remember exactly how far I got.

> 
> I assume you tested 720x480Under and 720x480Fit NTSC modes to be working 
> fine for you. Am I right ?

I tested these modes, for the patch on my website, Under worked in 16
bit color and not 24 bit, Fit did not work at all, Over sorta worked at
16 bit color, but shifted down a bit.  For the attached patch, it might
be better.

On another note, if you look at the latest via driver, they got rid of
the Fit resolutions entirely.  I'm wondering if we should do the same...
I was also wondering how hard it would be to rip out all the tvout code
in the openchrome driver and replace it with the tvout code from via's
driver.  since in history openchrome was a fork off that driver in the
first place, it might not be too difficult, and, we already know that
via's driver supports more hardware than openchrome does.  Is it worth
doing?

> 720x756Over PAL is confirmed to be working by Roy. Hopefully Udo will 
> also confirm.
> 
> What about 720x480P and 640x480 ?

timmy said he was using the old 720x480Under successfully as a
progressive scan mode, and that is what the 480P one is, so I assume
that timmy would be the one who could tell if it still works the same as
he expects.

the 640x480 PAL and NTSC modes were tested and found to not be working.


Hope that helps

Ken

> 
> Regards,
> Xavier
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tmpdiff
Type: text/x-patch
Size: 12396 bytes
Desc: not available
Url : http://wiki.openchrome.org/pipermail/openchrome-users/attachments/20070203/a86d3c62/tmpdiff-0001.bin